Table 2.

Optimal weights learned for the simulated shapes with mean and lowest NDCG over all K = 1..143, for various references standards

RS1: the inverse of the absolute difference of compactness for both shapes. RS2–RS5: the inverse of a weighted sum of squared differences between the corresponding parameters (lobulation amplitude, lobulation frequency, and eccentricity) that were used to generate each shape, where for (RS2) all weights = 1. RS3: the weights were 3, 1, and 1; RS4: the weights were 1, 3, and 1; and RS5: the weights were 1, 1, and 3, for eccentricity, lobulation amplitude, and lobulation frequency, respectively. RS6: the inverse of the absolute difference in the axis ratios between ellipses fitted to each shape